We were actively chatting with Justine yesterday, seems like the implementation is at least 2x faster than tinyBLAS on her workstation. The whole discussion is

We were actively chatting with Justine yesterday, seems like the implementation is at least 2x faster than tinyBLAS on her workstation. The whole discussion is in Mozilla AI discord: https://discord.com/invite/NSnjHmT5xY https://discord.com/invite/NSnjHmT5xY